Tinker by Thinking Machines
Hacker ApprovedFine-tuning platform in AI tools
Fine-tuning service that exposes training primitives; people find it smooth for large RL workflows. Critics say costs can be high.
Fine-tuning service that exposes training primitives; people find it smooth for large RL workflows. Critics say costs can be high.
16 people gave an opinion
- recommend it
- 13
- criticize it
- 3
- approval share
- 81%
Recommended in 5 of 13 months
- November 2025: recommended
- December 2025: not recommended
- January 2026: not recommended
- February 2026: recommended
- March 2026: not recommended
- April 2026: not recommended
- May 2026: not recommended
- June 2026: not recommended
- July 2026: recommended
- August 2026: recommended
- September 2026: not recommended
- October 2026: recommended
- November 2026: not recommended (month in progress)
Why it carries the badge
- 13 people recommend itNeeds 10 or more, met.
- 4.3 recommenders for every criticNeeds 2 to 1 or better, met.
- Recommended in 5 of 13 monthsNeeds 4 or more, met.
The same three rules apply to every product, and the badge can’t be bought. How the badge works
What people say about it
The reasons people give, with the number of comments that give each one. Open a reason to read what they said.
Why people recommend it
Each mark is one comment
Smooth for large RL workflows
“It's one of the smoother experiences we've had running large scale RL workflows.”
Comment from 4 Nov 2025 on Hacker News “RL-forms of distillation with Tinker! I hope that helps.”
Comment from 26 Oct 2026 in “Ask HN: Who is doing distillation or fine tuning?” Good docs and starting point
“Unsloth and Tinker (from Mira's TML) are great places to start”
Comment from 10 Aug 2026 in “Meta Muse Glimmer – open weights 30B local coding model” “They have good docs on finetuning in general here”
Comment from 16 Jul 2026 in “Inkling: Our Open-Weights Model” Strong fit for open base models
“Open base models that can be fine tuned on Tinker is a great business model IMO.”
Comment from 15 Jul 2026 in “Inkling: Our Open-Weights Model” “Open base models that can be fine tuned on Tinker is a great business model IMO”
Comment from 16 Jul 2026 in “Inkling: Our Open-Weights Model”
What people warn about
Each mark is one comment
Costs high; economics often don't work
“even with a finetuning-as-a-serice like Tinker (the one from Thinking Machines) which is pretty cheap, the economics didn't work out that well”
Comment from 20 Jun 2026 in “I built a WhatsApp clone of myself; a friend agreed on a password with it” “Expensive, in the thousands.”
Comment from 30 Jul 2026 in “Show HN: Distill and serve small models with frontier quality for half the cost”
What people said
14 of the 16 comments that mention it, newest first. Each links to the original comment.
“RL-forms of distillation with Tinker! I hope that helps.”
“The Tinker implementation is my favorite part, fine tuning is much better than a sea of context files.”
“Unsloth and Tinker (from Mira's TML) are great places to start”
“Expensive, in the thousands.”
“They have good docs on finetuning in general here”
“I found it easier and cheaper to just build from scratch”
“Open base models that can be fine tuned on Tinker is a great business model IMO”
“Open base models that can be fine tuned on Tinker is a great business model IMO.”
“its first-class integration with Tinker”
“even with a finetuning-as-a-serice like Tinker (the one from Thinking Machines) which is pretty cheap, the economics didn't work out that well”
“it exposes the primitives forward_backward, optim_step, sample and checkpoints”
“Made fine-tuning 30B parameter Qwen model feasible by reducing trainable parameters significantly”
“even though it feels unfinished, it's pretty great”
“It's one of the smoother experiences we've had running large scale RL workflows.”
Counts come from 1,795,640 Hacker News comments, November 2025 to November 2026. How we count. Report a mistake.